When electricity first emerged as a source of reliable power for industry, initially factories that had used mechanical sources of power retrofitted so that the complicated contraptions made of belts cams and axles that would allow motion to be distributed were powered by an electric motor. It took many years for the ways of working and the infrastructure to catch up to the possibility of putting electric motors in the machines and dispensing with the Heath Robinson nonsense.
In the middle of the last century as the efficiency of diesel fuel became increasingly attractive, a number of train locomotive manufacturers dabbled with the idea of building diesel-powered steam trains. In hindsight it’s a daft idea, but at the time the opportunity to be able to improve efficiency without wholesale infrastructure change must have seemed at least fleetingly attractive.
When personal computers first emerged, the killer business apps were based on digital manifestations of the typewriter, the slide projector and the Venetian system of accounting. The first killer network app used the metaphor of the postal system, and featured icons of envelopes at every given opportunity. Three decades on and at least three of those four ancient metaphors still rule in most businesses.
It’s really hard for a new technology to emerge and for it to find it’s new niche. Often there are long periods between first adoption and the new technology really finding its place, doing something that you simply couldn’t do before. Sometimes all they ever do is to ape the past.
In the recent hubbub about AI, and in particular the OpenAI ChatGPT engine, I’ve been thinking a great deal about how there is huge opportunity for us to automate things in a way that is simply analogous to how we currently do things. Unlike the electric motor or the word processor, however, I fear that this will result in the automation and scaling of stupid things.
Take, for example, performance management processes. I have heard folk talking about how ChatGPT could take the hard work out of creating performance objectives and even performance reviews. I can already smell the startups honing their propositions for Automated Performance Review, powered by AI.
On the face of it, what a wonderful idea. Nobody likes writing the bloody things, so why not hand off the drudgery to the machines? Well, because the whole process of performance management is the issue, not the filling out of interminable forms. Stick an NLP bot in the middle and you’ve created machines to do the work of communicating between humans. If no one can be bothered with performance management, why not get rid of performance management, rather than sticking daft exercises in the path of robots?
There is so much prose written to support processes and activities that are stupid. Procurement tenders, content marketing and CVs are three more that immediately spring to mind.
If you’ve identified a use case for removing the need for humans to write prose, can I boldly suggest that you might have identified a use case to simply stop doing? For goodness sake don’t give that work to machines. In the short term you’ll simply burn money for the sake of not looking at a problem properly, and in the long term they’ll remember and hold all that drudgery against you when they do become sentient.
mmitII 